Gray matter volume alterations in patients with strabismus and amblyopia: voxel-based morphometry study

This study proposes the use of the voxel-based morphometry (VBM) technique to investigate structural alterations of the cerebral cortex in patients with strabismus and amblyopia (SA). Sixteen patients with SA and sixteen healthy controls (HCs) underwent magnetic resonance imaging. Original whole brain images were analyzed using the VBM method. Pearson correlation analysis was performed to evaluate the relationship between mean gray matter volume (GMV) and clinical manifestations. Receiver operating characteristic (ROC) curve analysis was applied to classify the mean GMV values of the SA group and HCs. Compared with the HCs, GMV values in the SA group showed a significant difference in the right superior temporal gyrus, posterior and anterior lobes of the cerebellum, bilateral parahippocampal gyrus, and left anterior cingulate cortex. The mean GMV value in the right superior temporal gyrus, posterior and anterior lobes of the cerebellum, and bilateral parahippocampal gyrus were negatively correlated with the angle of strabismus. The ROC curve analysis of each cerebral region confirmed the accuracy of the area under the curve. Patients with SA have reduced GMV values in some brain regions. These findings might help to reveal the potential pathogenesis of SA and its relationship with the atrophy of specific regions of the brain.

Strabismus is a common visual developmental disease due to dysfunction of the extraocular muscles, which is thought to be associated with a deformity in the cerebral visual pathway that intermediates ocular movement 1 , and leads to the lesion of stereopsis and binocularity 2 . The prevalence of adult new-onset strabismus is reportedly 54.1 cases per 100,000 people 3 , and strabismus is often correlated with amblyopia 4 . Amblyopia is an optic disorder that is defined as a decline in visual acuity (VA) resulting from abnormal binocular development 5 . Amblyopia is likely to contribute to perceptual strabismus, and, conversely, strabismus might exacerbate amblyopia.
In recent years, magnetic resonance imaging (MRI) has been increasingly developed, and offers a noninvasive imaging approach to determine both functional and structural alterations in the cerebrum 6 . Some investigators have used various MRI analytical techniques to detect intrinsic activation of brain regions in amblyopic and strabismic patients, such as amplitude of low-frequency fluctuation (ALFF), degree centrality (DC), and regional homogeneity (ReHo) These techniques facilitate the investigation of spontaneous activities of the brain and reveal the underlying mechanisms occurring in the diseased eye [7][8][9] . In our previous studies, we used the ALFF method to explore alterations in cerebral activity in participants with strabismus and amblyopia (SA) 10 . We observed functional cerebral alterations in subjects with SA; however, morphological differences in the neuromechanism in the brain of patients with SA remained unclear.
Voxel-based morphometry (VBM) is a broadly applied technique used to analyze cerebral alterations at the voxel level, which can reveal the anatomical changes in whole brain tissue by quantitatively evaluating the density, volume, and distribution of gray matter and white matter across several cerebral regions 11 . Recently, the VBM approach has been widely utilized for the diagnostic, therapeutic, and prognostic assessment of many ocular diseases, such as cataract, glaucoma, and optic neuritis [12][13][14] . The technique has been successfully proven to be reliable and credible in identifying abnormal anatomy in ophthalmological disorders and conducive to elucidate MRI parameters. All participants were scanned with a 3-Tesla MRI scanner (Trio, Siemens, Munich, Germany). High-resolution T1-weighted images with a magnetization-prepared rapid gradient echo (MP-RAGE) sequence was obtained. The following comprised the specific scanning parameters: 176 images of 1 mm section thickness; echo time = 2.26 ms; repetition time = 1900 ms; field of view = 215 × 230 mm; acquisition matrix = 256 × 256; flip angle = 9°. VBM analysis. We preprocessed the structural image data using the voxel-based morphometry toolbox (VBM8, http:// dbm. neuro. uni-jena. de/ vbm8/), Statistical Parametric Mapping (SPM8, http:// www. fil. ion. ucl. ac. uk), and MATLAB 7.9.0 software (R2009b; The Mathworks, Inc, Natick, MA, USA). The brains were divided into three parts as gray matter, white matter and cerebrospinal fluid using the default estimation options on VBM8, as previously described 18 . The preprocessed data were then standardized to conform to the Montreal Neurological Institute (MNI) criteria. The generated template produced by DARTEL (Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra) analysis was used to standardize the white matter and gray matter of subjects. The modulated volumes were then smoothed with a Gaussian kernel of 6-mm full width at half maximum.
Statistical analysis. The SPSS 20.0 software (SPSS, IBM Corporation, NY, USA) was used to compare the differences in clinical data between patients with SA and HCs. In addition, the independent sample t-test (age and best-corrected VA) and chi-squared test (sex) were applied. A p value < 0.05 was set as statistically significant.
The differences in gray matter volume (GMV) and white matter volume (WMV) between the SA and HC groups were analyzed with the SPM8 toolkit using general linear model (GLM) analysis. The level of significance was considered to be p < 0.01.
Receiver operating characteristic (ROC) curve analysis was applied to distinguish the mean GMV values in different cerebral areas of the patients with SA from those of HCs.
Brain-behavior analysis. The mean GMV and WMV values were calculated by averaging every GMV or WMV value over entire voxels for each regions of difference (RODs) according to the VBM measuring results, using the Resting-State fMRI Data Analysis Toolkit (REST) (http:// www. restf mri. net). Correlation analysis was applied to explore the relationship between the mean GMV value of specific cerebral regions and clinical manifestations in the SA group. A p value < 0.01 was deemed to indicate a statistically significant difference. Ethics approval. This study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Nanchang University. The procedures were in accordance with the principles of the Declaration of Helsinki.
Informed consent. Informed consent was obtained from all individual participants included in the study, with awareness of the study purposes and potential risks.

Results
Demographics and visual measurements. No statistical differences were noted in age (p = 0.626) or best-corrected VA of the fellow eye (p = 0.847) between subjects with SA and HCs. The best-corrected VA of the amblyopic eye in the SA group was significantly higher than that in the HC group (p < 0.001) (details are presented in Tables 1). VBM differences. Compared with the HCs, GMV values in the SA group showed significant reduction in the right superior temporal gyrus, posterior and anterior lobes of the cerebellum, bilateral parahippocampal  1 and Table 2). However, there were no statistically significant differences in mean values of whole brain GMV and WMV between two groups (Table 3).  Table 4.

Discussion
Strabismus and amblyopia are common visual developmental diseases. Their underlying pathogenesis and the relationship and interaction between these two conditions have thus attracted much attention. Previous reports have illustrated that early aberrant visual experience could obstruct interocular alignment and ocular movement, thereby leading to strabismus 19 . In addition, disturbance of the development of the sensory and visual cortex might result in amblyopia 20 . Therefore, the presence of either one of these disorders in early childhood could probably lead to the other (Fig. 3). Analyses based on VBM constitute a reliable MRI method that can reveal cortical alterations and explore morphological changes in gray matter and white matter in the brain. Previous studies have applied the technique in other ophthalmological diseases (Table 5). To our best knowledge, the present research is the very first to apply the VBM method to patients with SA.
In this study, we discovered a significantly reduced GMV in the right superior temporal gyrus, posterior and anterior lobes of the cerebellum, bilateral parahippocampal gyrus, and left anterior cingulate cortex, with impaired visual function in patients with SA (Fig. 4).
The cerebellum is an isolated structure at the bottom of the cerebrum, tucked beneath the cerebral hemispheres. Previous studies have reported that the cerebellum is involved in motion control and perception 21 , especially the execution of ocular movements [22][23][24] and visually guided saccades 25 . Ouyang et al. reported that individuals with concomitant strabismus had declining GMV in the left posterior lobe of the cerebellum 16 . Studies on individuals with concomitant exotropia have shown that the DC value and mean diffusivity value in the right posterior lobe of the cerebellum were markedly reduced 26,27 . Furthermore, our previous research on patients with SA demonstrated that the ALFF values in the left posterior lobe of the cerebellum were reduced, when compared with HCs 10 . Similar alterations were observed in subjects with anisometropic amblyopia and reduced spontaneous activity in the cerebellum 28 . In line with previous reports, we also observed that subjects with SA showed significantly lower GMV values in the right posterior lobe and anterior lobe of the cerebellum, indicating damage associated with the execution of eye movement in this region.
The superior temporal gyrus lies laterally to the cerebrum, and is located in a superior position to the external ear. It is implicated in auditory processing, language comprehension, as well as social cognition 29,30 . Recent investigations on the function of the superior temporal gyrus have uncovered its relationship with eye movement processing and visual analyses of social information transmitted by gaze 11,31 . Prior studies have reported inconsistent results in this area of the brain. Some have demonstrated that adults with concomitant exotropia exhibit      Figure 3. Relationship between MRI images and strabismus with amblyopia. Notes: Amblyopia is strongly correlated with the presence of strabismus during early childhood. Early aberrant visual experience could obstruct interocular alignment and ocular movement, leading to strabismus, and disturbing the sensory and visual cortex development might result in amblyopia. www.nature.com/scientificreports/ subjects with SA showed decreased GMV values in the right superior temporal gyrus, which further corroborates the notion that SA may lead to dysfunction of the temporal gyrus. The parahippocampal gyrus is a GM cortical area of the brain that encircles the hippocampus, and plays a major role in memory encoding and retrieval. A previous report found reduced GMV values in the bilateral parahippocampal gyrus of patients with monocular amblyopia 15 . Similarly, reduced gray matter density of the parahippocampal gyrus was detected in children with amblyopia 17 . In addition, Berberat and associates also proved that images of strabismus could induce activation of the parahippocampal gyrus, when compared with images of normal eyes 33 . Consistent with these findings, the reduced GMV value of the bilateral parahippocampal gyrus in the present study may indicate injury in patients with SA.
The cingulate gyrus is an essential portion of the limbic system, which plays a role in the formation of emotions 34 , depression 35 , and pain 36 . The limbic system is closely associated with memory and emotion 37 . Previous studies about patients with concomitant strabismus have shown activation in the region of the anterior cingulate cortex 27,38 . Adult subjects with concomitant esotropia also evidently exhibit increased voxel-wise degree centrality values in the bilateral anterior cingulate cortex 26 . Moreover, Chan et al. showed that adults with strabismus exhibit greater gray matter volume in the anterior cingulate gyrus 4 . However, reduced GMV was observed in the left anterior cingulate cortex of individuals with optic neuritis 13 . In the present study, GMV reduction might have been attributed mainly to the poor vision associated with amblyopia, rather than strabismus.
Correlation analysis revealed that the angle of strabismus in SA is negatively correlated with the mean GMV values of several brain regions (Fig. 2). A greater angle of strabismus indicated more severe SA. Thus, we might presume that more severe strabismus can lead to greater atrophy of the gray matter. Accordingly, we suspect the reduction in GMV in those cerebral areas may be the underlying pathological mechanism of ocular motor disorders in patients with SA.
The ROC curve analyses were used to distinguish diseased individuals from HCs, with high sensitivity and specificity (Fig. 2F). When the AUC was over 0.8, it denoted perfect accuracy; an AUC between 0.6 and 0.8 meant the accuracy was moderate; and if the AUC was less than 0.6, diagnostic value was limited. In the present study, ROC curve analysis revealed that perfect AUC values were observed among all areas of interest, including the superior temporal gyrus, posterior and anterior lobes of the cerebellum, parahippocampal gyrus, and anterior cingulate cortex, indicating that the VBM method could be useful in characterizing the neural mechanisms underlying SA, and may be capable of detecting early biomarkers of SA.
Nevertheless, some limitations in the present research need to be considered. For instance, a relatively insufficient number of participants were enrolled, which may have affected reliability. In addition, various types of strabismus were implicated, which are supposed to be categorized in a subsequent study, to probe the structural alterations of the brain more precisely.

Conclusions
VBM is a computational approach to neuroanatomy that allows for comprehensive measurement of cerebral differences, not just in specific structures with traditional morphometry, but throughout the entire brain. In this study, we found changes not only in the visual-related and motion-related brain regions, which we could predict, but also in emotion-related area (cingulate gyrus). These findings might help to reveal the potential pathogenesis of SA and its relationship with atrophy in specific regions of the brain.